BPS2026 – Constructing IDP conformational ensembles with experimental data through multi-replica averaged simulation

Intrinsically disordered proteins (IDPs) are highly dynamic, yet their conformational ensembles encode critical biological functions. Capturing these ensembles remains a central challenge due to their structural plasticity, the limitations of current models, and the diversity of experimental restraints. We introduce multi-replica averaged restraint simulation (MARS), a biased simulation framework that integrates heterogeneous experimental data by applying restraints on observables averaged across multiple parallel replicas. This approach enables the construction of structural ensembles that are consistent with both global and local measurements. We applied MARS to the N-terminal domain of estrogen receptor α, incorporating small-angle X-ray scattering (SAXS) data and paramagnetic relaxation enhancement (PRE) measurements at six labeling sites. The resulting ensembles reproduced the experimental data with high fidelity and were independently validated using NMR relaxation parameters. Beyond data reproduction, MARS uncovered transient intra-domain interactions consistent with mutagenesis studies, revealing structural features that were not apparent from the raw experimental data alone. This framework broadens the toolkit for IDP research by providing a generalizable strategy to integrate diverse biophysical measurements. By resolving complex conformational landscapes, MARS offers new opportunities to understand how sequence encodes function in IDPs.